Skip to main content

3D shape and material properties for recognition

Objective

Investigation of the neural selectivity in key regions (areas TE and V4) of the ventral visual pathway of primates for pictorial 3D shape cues and material properties. Comparison of this neuronal coding with the characteristics of human perception of 3D shape defined by pictorial cues and materials. Provide theoretical foundation for processing of pictorial cues and material properties in computers. Implement the theory and compare computer performance with human perception and neuronal coding.

OBJECTIVES
Restore the cross fertilization between biological vision (neuroscience and psychophysics) and computer vision. Study the coding of pictorial cues for 3D shape and of material properties in areas TE and V4. Search for grouping of properties in either area and relate to connectivity. Study the perception of 3D shape defined by pictorial cues and of material properties. Develop mathematical theories necessary for implementation of pictorial cue and material processing in computers. Implementation of the theory to create a flexible system.

DESCRIPTION OF WORK
Study the selectivity of TE and V4 neurons for texture defined (single and double curved) surfaces.
Study selectivity of TE and V4 neurons for shaded stimuli derived from theory and perceptual studies.
Study coding of material properties in V4 and TE.
Study invariance of the selectivity for parsing, changes in illumination and elimination of cues.
Map functionally the cortical regions involved in material processing.
Study the perception of surfaces and objects defined by texture or combination of apparent contour and shading under a range of (illumination changes etc) conditions.
Study the perception of materials under different conditions of illumination, texture manipulation and elimination of cues.
Develop the mathematical theory of singularities in combinations of apparent contour and different types of shading.
Inverse the 3D to 2D projection and recover illuminated objects and sources of illumination using variational tools and study the classes of solutions for a given image.
Train computers to discriminate materials and extract processing principles which can be implemented. Integrate material processing with 3D geometry processing.

Funding Scheme

CSC - Cost-sharing contracts

Coordinator

KATHOLIEKE UNIVERSITEIT LEUVEN
Address
Oude Markt 13
3000 Leuven
Belgium

Participants (5)

INSTITUT NATIONAL DE RECHERCHE EN INFORMATIQUE ET EN AUTOMATIQUE
France
Address
Domaine De Voluceau
78153 Le Chesnay
KUNGLIGA TEKNISKA HOEGSKOLAN
Sweden
Address
Valhallavaegen 79
100 44 Stockholm
THE UNIVERSITY OF LIVERPOOL
United Kingdom
Address
Abercromby Square
L69 3BX Liverpool
UNIVERSITE PAUL SABATIER
France
Address
118, Route De Narbonne
31062 Toulouse
UNIVERSITEIT UTRECHT
Netherlands
Address
Heidelberglaan 8
3584 CS Utrecht